Full three-dimensonal reconstruction of the dyadic Green tensor from electron energy loss spectroscopy of plasmonic nanoparticles

Electron energy loss spectroscopy (EELS) has emerged as a powerful tool for the investigation of plasmonic nanoparticles, but the interpretation of EELS results in in terms of optical quantities, such as the photonic local density of states, remains challenging. Recent work has demonstrated that under restrictive assumptions, including the applicability of the quasistatic approximation and a plasmonic response governed by a single mode, one can rephrase EELS as a tomography scheme for the reconstruction of plasmonic eigenmodes. In this paper we lift these restrictions by formulating EELS as an inverse problem, and show that the complete dyadic Green tensor can be reconstructed for plasmonic particles of arbitrary shape. The key steps underlying our approach are a generic singular value decomposition of the dyadic Green tensor and a compressed sensing optimization for the determination of the expansion coefficients. We demonstrate the applicability of our scheme for prototypical nanorod, bowtie, and cube geometries.